Here's another one.
Assumptions:
I'm assuming my 0.1 procedure is actually targeting 0.1 accurately. That is the reference and thus the average of all the 0.1's is taken as the target "pixel volume". Note the actual estimate for the z-diameter (20% of the XY diameter in this case, i.e., something like a hamburger bun shape) does not affect the final normalized numbers but it's used just as a number to plug into the volume formula, 4/3 * pi * r1 * r2 *r3.
Good news:
The 0.1 procedure is pretty nicely repeatable, ranging from 88% from 108% of mean. On the previous attempt it ranged from 80% to 120% which is still not terrible. This was done on 2 different syringes.
The really good news:
This round saw calculated a drop as 0.027. Last round saw a drop as 0.024. The two rounds were done on two different syringes. If we take as given that the drop procedure is very repeatable (which it should be the most), then this is good evidence that the 0.1 procedure is also repeatable.
The bad news:
The difference between "0.25" and "0.1" was not nearly what it should be.
How a full blown study would work:
We'd need 30 examples of 0.25, 0.1, and 1 drop. Within each dose category, each sample should be from a different syringe of the same model so that that variable can be accounted for. Ideally each sample would come from a different vial of Lantus, though that would probably not be feasible. Each time we'd use the drop as the reference because we can safely assume the drop procedure is dead solid. With this method you can do 3-4 samples on each photo as long as the photo also contained a drop as a reference. They need to be close together so that the camera angle is not a factor, and the photo has to be taken directly flat overhead.
We'd ideally want each person participating to provide 30 samples of each dose size. Then we'd calculate something called a capability index for to see how each person does.
What it would and wouldn't show:
It would NOT tell us what the actual dose is in units or CCs that we're administering when we say we're doing 0.25, 0.1, or a drop.
It WOULD tell us how consistent we're all being, and how the different doses compare to one another.
Any takers?